The history of the Interlisp Programming Environment

[Anders Gilbro Nielsen]

In my master thesis I need an appendix describing the history of the
Interlisp programming environment, but I'm not quite sure of what
versions have been around and what they actually contained of new
additions to the environment. But here is the history (in short) as I see it.
It's supposed to end up in a time-line schematic shape. I probably
got a lot of things wrong, and I got some questions too. So if anyone can
help my it would be much appreciated.

History of Interlisp:

The Interlisp programming environment was originally introduced as a text-
based environment, which is described in article "The Interlisp Programming
Environment" by Warren Teitelman and Larry Masinter, Computer, 14:4, 1981.

Later on a graphical user interface was added to the environment, which made
especially the structure editor easier to use. The environmet was at this
point truly incremental since any changes made by the structure editor was
made directly in the internal program describtion. The system - now named
Interlisp-D - is described by B.A. Sheil in article "Power Tools for Pro-
grammers", Datamation Magazine, 1983.

There were a couple of Interlisp-D versions around. For example the Koto
release and the Lyrics release. In the latest version CommonLisp was
integrated, which ment that the editor - SEdit - not anymore was a
structure editor. Now the Lisp-code was parsed following an editsession,
giving almost the same functionality as the structure editor. All other tools
were the same.

Now the questions are:

Am I wrong somewhere?

When and in which version was CommonLisp integrated?

What was the difference between the Koto and Lyrics Release?

When was the graphical user interface added and what is the name of the
interface?

What is happening to the Interlisp Environment these days. Is annything
similar coming up?

Any references to articles containing the answers are also welcome.
Please e-mail or repply to the newslist.

Thank you.

Anders Gilbro Nielsen
Roskilde University Centre
Computer Science Dept.
House 20.1
4000 Roskilde, Denmark

E-mail: ba...@gorm.ruc.dk

PS. I regret any spelling errors. I'm not used to write in English.

[stic...@cps.msu.edu]

if you are writing the "history of interlisp" in just several paragraphs,
then first, you have a very difficult task.

second, without comment on what you did include, i will suggest that you
need an addition. the LOOPS addition to interlispD was in fact very
influential within the D community. there is an artile in AI Mag in 82 or
83 (i think) by stefik, bobrow, and mital that is about LOOPS.

---jon---

------------------------------------------------------------------
Jon Sticklen tele: +517-353-3711
AI/KBS Lab - Comp Sci Dept FAX: +517-336-1061
A714 Wells Hall office in A392 Eng Bldg
Michigan State University
East Lansing, MI 48824-1027 home telephone: +517-337-5690

[Bob Bane]

I've only hacked Interlisp since 1985, but I was there when Lyric was
implemented. The major difference between it and Koto was Common Lisp;
the most visible change was the addition of packages.

Your description of SEdit is half-right; SEdit IS a structure editor,
but Common Lisp semantics require a structure mung between the code you
edit and the code that is executed (primarily comment-stripping). If
you REALLY want to, you can use pure Common Lisp text files in post-Koto
Interlisp.

These days, the environment is named Medley, is sold by Venue, and is
actually being sold profitably for the first time in its existence. It
has been ported to many Un*x workstations via an emulator that runs the
same compiled-code format as the D-machines. Possible future extensions
include making the Interlisp window system an interface to X windows,
and ports to non-Un*x platforms.

Your English is excellent; MUCH better than my Danish would be under the
circumstances. Could you send Venue a copy of your thesis when it's
finished? Thanks!

Venue
1549 Industrial Road
San Carlos CA 94070
USA

- Bob Bane

[Michel_Den...@xerox.com]

"There were a couple of Interlisp-D versions around. For example the
Koto release and the Lyrics release"

More than a couple. At first, new releases had no formal names - they were
simply announced via mail messages. For example:

-------

Date: 24 FEB 1982 2139-PST
From: MASINTER.PA
Subject: Newer version of Interlisp-D available
To: LispUsers^

A slightly newer version of the Interlisp-D system has been stored on file
servers Phylum, Indigo, Rose and Cask, including, on the directory
<Lisp>Current>, the files

Lisp.run, Lisp.sysout, DolphinLispMC.Run.

-------

I believe the first named release was called "ConBrio", dating to mid-1982.
This was followed by "Wind", which was a pre-release version of "Chorus", which
was released 1/13/83. Chrous recoded most of the low-level BCPL code in Lisp.
The window package was completely revised and floating point moved to
microcode.

The next release was Fugue.0 (6/30/83). Fugue.1 was never released. Fugue.2
came out 8/18/83, and Fugue.3 came out 10/13/83. Fugue marked the addition of
multi-processing and NS support. Chat and DEdit were promoted from Lispusers
to the base system.

The next major release was Carol, on Jan. 27, 1984. This included a local file
system for Dandelions. Carol.1 came out 3/2/84. The official Carol
announcement was June 21, 1984.

Next came Intermezzo, in early 1985. Intermezzo included a new TEdit format,
NS Chat, and expanded vmem to 32 meg. By this point, the documentation was
starting to look "official". Koto came out later in 1985 and included the
first edition of the integrated IRM (until then, the documentation was the old
Interlisp-10 manual of Oct. 1978, plus a "D" supplement.

Lyric came out in early 1987 and had major changes, including a compiler
incompatibility. Compiled files were renamed ".LCOM" from ".DCOM". Fonts and
streams underwent major revisions too in Lyric. Common Lisp was included.
SEdit was included (although it was available as a Lispusers package for Koto).
The language was now called "Xerox Lisp".

Around that time, Xerox spun off the Lisp group into the ill-fated Envos Corp.,
which released Medley. Medley 1.1 came out in January 1989, 1.2 last year, and
2.0 just a few months ago. Medley was the first release to run on Suns. I
understand that there will be no more new names, just incremented version
numbers.

"When was the graphical user interface added and what is the name of
the interface?"

I believe this goes clear back to the very beginning of time, i.e. the
legendary Alto Lisp (I tried loading it once, but couldn't get it to run.
Larry Masinter tells me it did work, although it had performance problems - not
too surprising considering the Alto-II had only 256K memory and 4 meg. of
disk). The basic window system as it exists today was certainly all there by
1982 when the 1100 (Dolphin) came out. The graphic interface is basic to the
system - I don't think it has a separate name. In the Sept. 1982 Interlisp-D
User's Guide, it's simply called "graphics display and window system".

- Michel

[Lennart Lovstrand]

Michel,

Great summary! Correct me if I'm wrong, but wasn't there also a
Harmony version between Carol and Intermezzo? I think TEdit was
pretty new then (and very buggy -- it was hard to believe that it
would ever be a serious editor at that time). I also seem to
remember an internal release in time for IJCAI in 1985 called "JCAI"
to fill in for the missing J version.

Then there was also BBN Lisp and/or Interlisp-10 for the DECsystem-10
and 20 and even a version for the IBM 360. I can't remember if they
had any version numbers or names, though. In any case, that was
before Interlisp became a Xerox product.

DEdit was the structure editor of choice before SEdit. It was
modelled after a TTY version that existed in Interlisp-10. It had
commands like "remove the 3rd element from the 2nd subexpression" and
"go up to levels and then prettyprint the current expression" -- all
with numbers and [mostly] single letter commands. And yes, it was
actually really useful if you didn't have a screen based editor.
Took a bit of getting used to, though.

--Lennart

[Michel_Den...@xerox.com]

"the Alto removable disk packs were all 2.5 MB capacity"

Sorry, I should have been more explicit - I was assuming a dual-drive system
(as most were by that point). I think each pack held around 2 meg., formatted.

- Michel

[Michel_Den...@xerox.com]

"wasn't there also a Harmony version between Carol and Intermezzo?"

There was indeed! Thanks for jogging my memory. My documentation is
mysteriously missing the Harmony announcement.

"DEdit was ... actually really useful if you didn't have a screen
based editor"

Well, DEdit of course *was* "screen-based". (It's still in the IRM - p. 16.1)
It had the same auto-formatting, point & click parenthesis-matching as SEdit.
It's command menu had items like After, Before, Delete, Replace, Switch, ( ), (
) out, etc. Nostalgia buffs who want the single-letter commands can still call
the function (EDITMODE 'TELETYPE) even in Medley.

- Michel

[Lennart Lovstrand]

> "DEdit was ... actually really useful if you didn't have a screen
> based editor"

Sorry, I wasn't very clear. I meant that the TTY based structure
editor was surprisingly usable on "non-D Interlisp" systems, even (or
especially ;-) compared with MacLisp that used Emacs in a different
process as its editor.

--Lennart

[Stan Lanning]

The early Interlisp manuals includes a brief history of Interlisp.
Here's some of the early versions:

Dec PDP-1 Lisp
D. G. Bobrow and D. L. Murphy at BBN
1966

SDS 940 Lisp
Bobrow and Murphy at BBN
1967
"This system contained the seeds for many of the capability and
features of the current systems: a compatible compiler and
interpreter, uniform error handling, an on-line LISP oriented editor,
sophisticated debugging facilities, etc. 940 LISP was also the first
LISP system to demonstrate the feasibility of using software paging
techniques and a large virtual memory in conjuntion with
list-processing system."

"BBN LISP" for the PDP-10
Bobrow, Murphy, A. K. Hartley, and W. Teitelman at BBN
1970
"The name of the system was changed from BBN-LISP to Interlisp in
1973, when the maintenance and development of the system evolved into
a joint effort between Bolt Beranek and Newman, and Xerox Palo Alto
Research Center."

By the way, the structure editor was not called DEdit -- DEdit was the
Display (hence the D) version of the TTY oriented structure editor.
From what I can tell, the TTY editor didn't have a name; it was refered
to as "the Interlisp editor".

There was also a VAX/VMS version of Interlisp, too, based (I think) or
the Interlisp Virtual Machine paper. (Is that what it was called? I
can't find my copy.) I used it for a while back in '84 (or was it '83?
or '82? Ah, how the memory plays tricks on you.). It came with a
LispUser package that turned the TTY based structure editor into a
screen based editor. Well, not quite -- the commands were the same old
structure editor commands, but at least you got a display of the
structure you were editing. That is, if you were running on a VT100.
--

-- smL

[Hans Koomen]

>> Date: 29 Oct 91 17:47:55 GMT
>> Subject: Re: The history of the Interlisp Programming Environment
>> To: info...@cis.ohio-state.edu
>> From: lan...@parc.xerox.com (Stan Lanning)
...

>> There was also a VAX/VMS version of Interlisp, too, based (I think) or
>> the Interlisp Virtual Machine paper. (Is that what it was called? I
>> can't find my copy.) I used it for a while back in '84 (or was it '83?
>> or '82? Ah, how the memory plays tricks on you.). It came with a
>> LispUser package that turned the TTY based structure editor into a
>> screen based editor. Well, not quite -- the commands were the same old
>> structure editor commands, but at least you got a display of the
>> structure you were editing. That is, if you were running on a VT100.

There was, indeed. Interlisp-Vax was developed at USC-ISI around 80-82. It
was more or less :-) 100% compatible with Interlisp-10, and was developed to
get past the pdp-10 memory limitations. Vaxen were hot, those days, and
offered big virtual memory. Interlisp-Vax allowed for much larger programs
than the pdp-10 could support. On the other hand, Interlisp-Vax never quite
reached the speed of Interlisp-10, which at that point had been fine tuned for
about a decade. Interlisp-Vax was based on the Interlisp VM description,
with various corrections and additions to make it "real-life." The display
editor mentioned above was a neat little addition that gave a 24x80 window
into the structure editor. I believe several hundred sites used Interlisp-Vax
at some time or another. For further details, see "Implementation of
Interlisp on the VAX" by Bates, Dyer & Koomen, 1982 ACM Symp. on Lisp and
Functional Programming.

- - - /-/ a n s Paper: Johannes A. G. M. Koomen, Ph.D.
Department of Computer Science
Phone: (716) 245-5401 State University of New York
EMail: koo...@cs.geneseo.edu Geneseo, NY 14454

PS. I believe at one time there was also a version of Interlisp for the IBM
370, developed at Lynkoping and/or Weizmann Institute, but I don't remember
much about it...

[Ralph P. Sobek]

Hans Koomen <koo...@brastias.cs.geneseo.edu> wrote:
| PS. I believe at one time there was also a version of Interlisp for the IBM
| 370, developed at Lynkoping and/or Weizmann Institute, but I don't remember
| much about it...

As an old user of Interlisp-10 (around '75) , I switched over to
Interlisp/360 /370 back in '79. This version was available from Dr.
Klaus Appel, U.D.A.C., University of Uppsala, Uppsala, Sweden as well
as Dr. Martin Raim, Dept. of Applied Mathematics, Weizmann Institute,
Rehovot, Israel. It ran in the IBM TSO and VM/CMS conversational
modes.

There were 2 systems, The Basic System which is compatible
with the "BBN LISP Reference Manua"l, 1971, and the Advanced System
(from Weizmann Institute) which runs major subsystems of Interlisp-10
(such as CLISP and DWIM) as described in W.Teitelman, "The Interlisp
Reference Manual", Xerox Corp., 1975.

A private communication that I had, stated that Interlisp/370
was actually based on the 1974 version of Interlisp-10: the version
before the spaghetti stack was added to Interlisp. As such, the
sections of "The Interlisp Reference Manual" marked with the '+'
change bars were *not* implemented in Interlisp/370.

Section 22 of the manual, the Programmer's Assistant and
LISPX, were ported by J.W. Goodwin at Linko"ping University, Norway.

In addition to the Programmer's Assistant and LISPX, the
Weizmann Institute also ported QLISP to Interlisp/370. QLISP is SRI's
implementation of pattern-directed invocation based upon their
in-house QA4 system.

Even though this version of Interlisp had none of the
graphical features of the D machine versions, it was the fastest
version of Lisp available anywhere for quite a long time. As IBM came
out with new processors, my code would execute in half the time!
(There was an embargo at that time on the exportation of the Xerox
Dorado.) As workstations started coming out with implementations of
Common Lisp, Interlisp/370 held its own with respect to execution
speed.

I would be curious to know if it still exists anywhere.

Cheers,

Ralph P. Sobek Disclaimer: The above ruminations are my own.
ra...@laas.fr Addresses are ordered by importance.
ra...@laas.uucp, or ...!uunet!laas!ralph
If all else fails, try: so...@eclair.Berkeley.EDU
Telephone: (+33-)61-33-62-66 FAX-1: (+33-)61-33-64-55 FAX-2: (+33-)61-55-35-77

[Dave Newman]

I can verify that there was a version of Interlisp-D called Harmony,
and that there was an unofficial version called JCAI (mostly for internal
use I think) that was released for IJCAI in '86 or so.

However, regarding the predecessor for SEdit, it was XEdit that was the
first structure editor that was available for Interlisp-D, and it was
indeed a strictly visual editor. DEdit may have preceded XEdit, though
I only ever used XEdit to edit code (it was neat: it edited the code in
memory, and you never actually edited the code files).

Also, one important point: Interlisp was ported to the VAX and some other
machines fairly early on. I think that Interlisp-Vax was available from ISI
at USC in the middle '80s. To get the real scoop, one would need to get
Larry Masinter or someone who has been around Interlisp for a long time.

>>Dave

[Dave Newman]

oops. Got a little confused, I did. When I talked about XEdit,
I was confused about another computer system I used. DEdit was
indeed the structure editor that preceeded SEdit. However, I'm
pretty sure that there was a line-based editor that used text-based
commands rather than the mouse and a menu. I never had the pleasure
of using that one.

>>Dave

[Henry S. Thompson]

In article <91Oct29.070...@alpha.xerox.com> Michel_Den...@xerox.com writes:

> "When was the graphical user interface added and what is the name of
> the interface?"
>
> I believe this goes clear back to the very beginning of time, i.e.
> the legendary Alto Lisp (I tried loading it once, but couldn't get
> it to run. Larry Masinter tells me it did work, although it had
> performance problems - not too surprising considering the Alto-II
> had only 256K memory and 4 meg. of disk). The basic window system
> as it exists today was certainly all there by 1982 when the 1100
> (Dolphin) came out. The graphic interface is basic to the system -
> I don't think it has a separate name. In the Sept. 1982
> Interlisp-D User's Guide, it's simply called "graphics display and
> window system".
>
> - Michel

There was in fact a predecessor to Interlisp-D's graphical user
interface which was briefly used at Xerox PARC around the time Alto
Lisp was being developed. It was a hybrid, using the Alto as a
display server for an Interlisp-10 client running on Maxc, PARC's
PDP-10 clone. I used it, perhaps around 1976-7, but can't remember
what it was called. Warren Teitelman built it, I think.
I think Bobrow and Goldstein's work which was the predecessor of Loops
used this at first.

While we're including attributions, the Alto Lisp effort was directed by
L. Peter Deutsch and the implementation mostly done by Willie Sue Haugeland.

ht
--
Henry Thompson, Human Communication Research Centre, University of Edinburgh
2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 31 650-4440
Fax: (44) 31 650-4587 ARPA: h...@cogsci.ed.ac.uk JANET: h...@uk.ac.ed.cogsci
UUCP: ...!uunet!mcsun!ukc!cogsci!ht